Abstract
This study addresses the challenges of infinite impulse response (IIR) system identification by introducing an improved cooperation search algorithm (ICSA). Improved cooperation search algorithm enhances the original cooperation search algorithm (CSA) through the integration of a pattern search algorithm and opposition-based learning, aiming to improve both exploration and exploitation capabilities. The algorithm's performance was evaluated against diverse IIR plants of varying orders using convergence analysis, scatter plots, and statistical metrics. Results demonstrate ICSA's superiority over CSA, achieving significantly lower mean squared error (MSE) values across different system orders and model types. Notably, ICSA outperformed CSA by up to 27 orders of magnitude for matched-order models and up to 95.85% for reduced-order models. The algorithm also exhibited more consistent performance, with substantially lower standard deviations in many cases. Statistical validation through the Wilcoxon signed-rank test further confirmed ICSA's enhanced performance. This research highlights ICSA's efficacy in producing efficient IIR systems, demonstrating its potential for more accurate system identification compared to existing methods.
Cite this article as: C. Ersalı and B. Hekimoğlu, "Improved cooperation search optimization algorithm for infinite impulse response system identification," Electrica, 24(3), 733-747, 2024.